How to bake half an object, but merge the whole object into the Paint Room?

[PolyPusher]

Sorry if this has been asked somewhere else, but I couldn't seem to find it if it is.  I've got a retopo mesh that is one half of what I'm trying to import into the Paint Room.  The UVs and geometry are going to be mirrored, so I only want to bake half of the object.  Yet, I want to merge the entire mesh into the Paint Room.  Is there some way to do this? Did I miss a checkbox somewhere that says "merge symmetrically after baking"?

 

The reason I ask is because if I mirror the geometry (using the Symmetry tool) before baking, I get some odd baking artifacts around 2 or 3 faces..  I know in other 3D programs (such as Maya), the process is to split/separate a symmetrical object before baking, then merge afterward. So my options currently are:

 

1. Bake only half the model, get perfect results texture-wise, but end up with only half a model in the paint room.

2. Bake the entire model, get a few texture artifacts (which could be smoothed out, but I shouldn't have to, right?)

 

I guess a workaround would be to export the whole retopo mesh as an OBJ, then reimport into the Paint Room, but that seems a bit clunky.

[digman]

First If possible share the 3DC file so others can look at it and to see if the artefact problem can be solved.

 

EDIT: I posted obj file when I meant the 3DC file.

Edited March 23, 2015 by digman

[PolyPusher]

Here's the 3DC file: https://www.dropbox.com/s/ea36sy5catllfoc/symmetry.3b?dl=0

[digman]

OK... I will give you how I fixed the baking... Your mileage made differ...

1. I first unwrapped the uv set, just in case something got messed up in your testings. Redid copying and pasting the identical uv islands to mirror them as you had done also.

That fixed most problems.

2. You had some large polygons meeting in a 5 point pole almost where the curve started again on the extruded cylinder. Poles, we have them in modelling and they are not bad, as they can be used to help guide the edge flow. They can become a problem causing artefacts though depending where they terminate. I added one extra polygon of course mirrored to the opposite axis and that fixed the final artefact in that area. I show in the picture the polygon I added.

 

The surface normals needed one more polygon to soften the transition at the pole.

Also I made sure I deleted the model in the paint room and the all the layers in the paint room except layer "0" which can not be deleted before rebaking.

You could very well know all the above but I others read this who might not be aware.

Version used: Linux 64 bit non cuda 4.5 Beta14.

Edited March 23, 2015 by digman

[PolyPusher]

Well, I see what you mean about the pole at the bottom, but I had similar areas that didn't result in a similar problem.  Also, why would it bake correctly with only half the mesh, but break as soon as it was mirrored (considering some of the problem areas were not on the edge of the mesh)?

 

Long-term, should I focus more on having "proper" geometry to avoid this problem?  I'm still not entirely sure what's the cause here.

[digman]

I retested, same result... I simply un - wrapped the model, copied and pasted the uv islands to mirror them. 3DC will overlay the copied uv to the exact other uv island, then baked. The pole error was there but the others were gone.

 

Mirror UV islands need to be perfect, maybe you somehow made one just a little different after you copied and pasted. I do not know of course.

After baking the retopo mesh the only error I found was the pole. All other areas were fixed.

 

The pole not only was at the place near the curve at the base of the cylinder but also you have the cylinder height curvature as well in the equation. Baking half the model removes the mirrors, hence no problems. I also just un-wrapped and did not mirror, baked, there were no errors as well. This cause me to think the problem was in your mirroring. In the midst of modeling, retopoing, seam creation, unwrapping and mirroring,  mistakes can happen... Been there done that.

 

Yes long term, it would be good to watch out for those kind of poles but if you find an artefact in the mesh just examine your retopo mesh to figure out what is going wrong then correct it.

 

The above is to the best of my knowledge and experience...

Edited March 24, 2015 by digman

[PolyPusher]

What's the process you used for mirroring the mesh?  I turned on Symmetry (with the 'S' key), then used the Symmetry command in the bottom of the toolbox.  That should be it, right?

[digman]

That is one method and the other is using virtual symmetry under the Retopo menu, then apply the symmetry. Next turn off virtual symmetry. Two methods, use one or the other depending upon the model and workflow.

 

Just as a test. I deleted half the model and then turned on symmetry. Used the symmetry command to mirror to the opposite X axis. Unwrapped and again used the copy and paste function to to overlay the uv islands on top of each other. Baked and no errors except the pole.. 

 

All I know is the problem was fixable and does not appear to be a bug. If similar problems keep happening then send a problem file to Andrew for him to look at through the Help menu inside of 3DC.

Edited March 24, 2015 by digman

[PolyPusher]

I've tried your method (which is really the same as I've been doing since the beginning), and I still seem to get the same problem.  Is this possibly a version issue?  You mentioned you were using the 4.5 beta, while I'm on 4.1.17.